My view called "Thematicization" for reference:
>> Thematicization is still too wide a pattern to be non-existent
>> since it would cover the development of the following:
>>
>> *-mn => *-mén-/-m&n- > *-mén-/*-mon-
>> *-tr => *-tér-/-t&r- > *-tér-/*-tor-
>> *-wr => *-wér-/-w&r- > *-wér-/*-wor-
>> *-l => *-él- /-&l- > *-él- /*-ol-
>> *-x => *-éx- /-&x- > *-éx- /*-ex-
Jens:
> You shouldn't call it that.
It's not "anaptyxis" since both athematic inanimate suffixes in
the leftmost column above and the derived animate counterparts next to
them coexisted by this stage. It really is most parallel with the
Thematicization seen elsewhere in later IE for the same purpose of
animatizing stems. Here, we are animatizing suffixes through a
similar process of adding a thematic vowel, the same thematic vowel
*& which we later find to alternate between *e and *o. So I'm sticking
with that name because it is the most appropriate.
> Is the last line as intended, i.e. not a typo for *-éx-/*-ox-?
No. We know that *h2 was unvoiced because *& necessarily becomes *e
before *-x in the animate collective that would become the feminine.
We see *-ex abundantly as the later feminine not **-ox. We've talked
about your **meg-o:x-s and the fact that it is completely unattested
(/mahaan/ and its paradigm fail to qualify because of unexplained -n).
I repeat: there can be no suffix **-ox- from this.
> The reduction of suffixal /e/ to /o/ is seen before all kinds
> of suffixal consonants, a few of which are voiceless, as in s-stems,
> where we have *H2áws-o:s 'dawn', and in the t-stem *nép-o:t-
> s 'nephew', with o-timbre despite the voicelessness.
There is no **nep- so that analysis of *nepot- is just silly. The
root as far as we're aware is indivisible and therefore completely
inadmissible here.
Of course, perhaps you'd be best to cite *men-ot- since segmentation
here _can_ be justified. Yet, what does this have to do with the above
pattern exactly? In order to use an animate *-ot- as counterproof of
the above pattern, we first need to find a correlating inanimate suffix
**-t. I'm not aware of such an _inanimate_ suffix, although a similar
looking suffix does form _animate_ stems like *nokW-t-s. Since it fails
the criteria of the pattern, *-ot- cannot be a product of Thematicization
nor can it be expected to be.
In all likelihood, stems in *-ot- are secondarily derived based on
the suffixes that did undergo Thematicization, so that a typically
found stem form like *men-t- was stretched to *men-ot- to conform
to the many other animate stems of the pattern CVC-VC-. Athematic
animate stems like *nokWt- were going the way of the do-do, as we all
know, since athematicism was growingly associated with inanimacy. We
have, of course, another way of animatizing this word: *men-t-o-. It's
merely a matter of where one chooses to place the thematic vowel to
make it more "animate-looking".
The 'dawn' stem is no doubt from *xeus-, but *-os- again shows that
final *-s was voiced as it would have been in the nominative xauso:s
based on your very ideas! Since the nominative *-s and the *s of the
suffix had merged, the *-s was in this unique case in _final_ position
in the nominative and thus _voiced_. Considering that insight, it shows
exactly the expected pattern as we find above:
eLIE *-és- /-&s- > *-és- /*-os-
Again, laryngeals in contrast are never affected by Final Voicing so
**-ox- cannot arise no matter what here, whether sheilded by an animate
nominative *-s or completely bare in later feminine nominatives.
> The rule distributing e/o according to the voicing properties of the
> following segment is valid only for stem-*final* vowels. And there
> it is independent of the accent.
Wrong, read above.
= gLeN